Cecil Moore, W5DXP wrote:
"Seems to me, the highest efficiency would be achieved by a source with
zero source impedance."

Me too, but zero source impedance does not match the load as required
for maximum power transfer. The best combination is then a source
impedance matching the load and which is also pracrically lossless. The
Class C amplifier does this by acting as a switch which is infinite in
impedance when open during a large part of the RF cycle and a near short
circuit to a low impedance (near zero Z) D.C. power source for the short
part of the RF cycle it is switched on. It is the time averaged
impedance which counts. Is this linear? No way, but the tank circuit is
able to remove the harmonics and turn current pulses into a low
distortion sine wave. Efficiency? Terman says on page 450 of his 1955
opus that Class C eddiciency is typically 60% to 80%. Compare that with
50% efficiency in a Class A amplifier.

Best regards, Richard Harrison, KB5WZI